The STG1702 supports an 16-bit pixel port (on pins P8 to P15), which the AT20C490 does not. As long as you don't select a 16-bit pixel configuration (in software), these pins are not used on the STG1702. The AT20c490 has inputs or voltage supply on all of those pins, so there should be clearly defined levels on all of these pins, and there will be no bus conflicts when swapping those chips. The AT20c490 has RS2 on one of the pins the STG1702 re-purposes for the 16-bit pixel port. This is not an issue because the ARK1000 BIOS access the DAC command register in a way that doesn't require RS2, so installing a DAC without RS2 input on an ARK1000 card will not cause any issues.

Thx for checking. What do you think about the STG1700J?
I found this, I hope it is a real datasheet

Great find. That's the complete data sheet. It will likely work. If you have trouble switching modes after you were in a true color mode, a BIOS patch might be helpful, but I expect no problems at all. Note this quote from the data sheet you attached: "On power up the magic access sequence is partially disabled through the default setting of Pixel Command Register, bit \[4\], so that the STG1700 is identified by existing video BIOS code as fast ATT20C49X". This is exactly what we need. The Pinout of the 1700 and 1702 is identical, so the comparison also applies to the 1700 - it should work.

Are the STG1700 and STG1702 RAMDACs better candidates for trying to force 1024x768x65k at 60 Hz on the ARK1000VL compared to ATT20C490-11?

feipoa wrote on 2023-11-22, 06:10:

Are the STG1700 and STG1702 RAMDACs better candidates for trying to force 1024x768x65k at 60 Hz on the ARK1000VL compared to ATT20C490-11?

I don't think so.
Tried the STG1700 on my card. It seems that it is not 100% compatible, I tried both the modified bios and the original Octec Bios:
1280x1024 256c 60 Hz => black screen.
800x600 24bit c => black screen on both 56Hz and 60Hz.
Even worse:
800x600 16 bit which worked before with 56Hz, 60Hz and 72Hz.
With the STG1700: 56Hz => OK
60Hz => some distortions
72Hz => picture completely broken.

Any ideas?

CoffeeOne wrote on 2023-11-19, 18:23:

Would you be so kind and test 1280x1024 with the Ark1491 RAMDAC if it works (longer than 10 seconds)?

I just retested this. I ran the ARK1491 at 1024x768x256c-60Hz with 80 MHz MCLK for over a minute. It is functional, but the image is not as clear as with the ATT20C490-11 RAMDAC. Even in pure DOS, the ATT RAMDAC is clearer than the ARK RAMDAC, so perhaps the image blurriness isn't solely due to VCLK, but is due to RAMDAC quality.

I have received the V53C104HP45 chips from utsource, however I suspect they are remarked.

Notice the streaking on the surface, which looks like someone sanded the surface to do the remark. Also, a 0635 datecode seems a bit too new for DIP FPM. What does the bottom of your V53C104HP45 look like? Curiously, the chips are indeed used, possibly even desoldered.

Next, I tried to use these chips in the ARK1000VL card, but they could not handle an MCLK at 80 MHz. The best they could handle was 70 MHz. My slower rated NPN modules could do 80 MHz without issue.

Were your V53C104HP45 chips able to handle 80 MHz?

feipoa wrote on 2023-12-04, 09:12:

I just retested this. I ran the ARK1491 at 1024x768x256c-60Hz with 80 MHz MCLK for over a minute. It is functional, but the imag […]
Show full quote

CoffeeOne wrote on 2023-11-19, 18:23:

Would you be so kind and test 1280x1024 with the Ark1491 RAMDAC if it works (longer than 10 seconds)?

I just retested this. I ran the ARK1491 at 1024x768x256c-60Hz with 80 MHz MCLK for over a minute. It is functional, but the image is not as clear as with the W82C490-11 RAMDAC. Even in pure DOS, the ATT RAMDAC is clearer than the ARK RAMDAC, so perhaps the image blurriness isn't solely due to VCLK, but is due to RAMDAC quality.

I have received the V53C104HP45 chips from utsource, however I suspect they are remarked.
V53C104HP45_ARK1000VL.JPG
V53C104HP45.JPG

Notice the streaking on the surface, which looks like someone sanded the surface to do the remark. Also, a 0635 datecode seems a bit too new for DIP FPM. What does the bottom of your V53C104HP45 look like? Curiously, the chips are indeed used, possibly even desoldered.

Next, I tried to use these chips in the ARK1000VL card, but they could not handle an MCLK at 80 MHz. The best they could handle was 70 MHz. My slower rated NPN modules could do 80 MHz without issue.

Were your V53C104HP45 chips able to handle 80 MHz?

I thought you have an ATT RAMDAC, but you wrote now W82C490-11(?)
I removed one of my 256kx4 RAM chips, the front side is clearly visible on the previous picture. They all have the same date code 9344.

With the original 70ns RAM chips I was able to clock at 65MHz (default setting was 60MHz).
Then with the Mosel Vitelic ones I jumped from 65MHz directly to 80MHz. Also tried 70MHz later on, that works, too as expected.

But back to the RAMDACs:
I would say the ST type I have now on the card is not compatible: For both 800x600 24bit and 1280x1024 8 bit, I have now a blank screen.
I found on Vogons some info about the ATT 20C491, there is no mentioning about 1280x1024, so I guess that is not compatible either.
So the only one compatible are the ATT 20C490 you have and the Winbond type. I am not 100% convinced about the Winbond chip.
Where did you get the ATT 20C490-11?

EDIT: Thank you for the test.
I now took a closer look at your RAMs, the main symbol of Mosel Vitelic is wrong on your chips, it is a triangle I think.
So I assume it are 256kx4 chips, but either 70ns or 60ns types.

Yes, I have ATT20C490-11. The W82 was a late night typo that I've since fixed.

Thanks for the rear end photo of your V53C104HP45. Our bottom markings looks similar, so I'm not sure what is going on with my chips not handling 80 MHz. You are correct about the V symbol being truncated at the bottom. That leads credence to them being counterfeit. Maybe I'll return them.

I bought ATT20C490-11 from utsource. Sometimes you get the genuine article, sometimes not.

That one seems to be available (easier?)
Bt481AKPJ110
https://www.datasheets360.com/part/detail/bt4 … 39902704947511/

CoffeeOne wrote on 2023-12-04, 21:57:

That one seems to be available (easier?)
Bt481AKPJ110
https://www.datasheets360.com/part/detail/bt4 … 39902704947511/

The BT481 is not software compatible. At least 64K modes will fail to work unless the BIOS is modified. I don't have hope that substituting this DAC will yield better result than substituting the ST1700 into the ARK card.

I am attempting to return my V53C104HP45 chips to utsource and get this peculiar reply from customer support.

1What we sent to you is original part, since you ordered used part, it is re-print and refurbished, so the symbol is a little different. Could you test the part? I believe they can work well.

Sounds like rubbish to me.

The parts might be repainted, but you got what the markings state.

rasz_pl wrote on 2023-12-05, 21:13:

The parts might be repainted, but you got what the markings state.

Why do you think that? What would be the purpose in remarking the chips with the same part number and speed?

Why would 50 ns NPN chips work at 80 MHz MCLK, but the 45 ns V53C chips would not?

CoffeeOne, were your original V53C chips able to work at 80 MHz?

feipoa wrote on 2023-12-05, 10:40:

I am attempting to return my V53C104HP45 chips to utsource and get this peculiar reply from customer support. […]
Show full quote

I am attempting to return my V53C104HP45 chips to utsource and get this peculiar reply from customer support.

Copy code to clipboard

1What we sent to you is original part, since you ordered used part, it is re-print and refurbished, so the symbol is a little different. Could you test the part? I believe they can work well.

Sounds like rubbish to me.

Yes, more than ridiculous.

feipoa wrote on 2023-12-05, 21:27:

Why do you think that? What would be the purpose in remarking the chips with the same part number and speed? […]
Show full quote

rasz_pl wrote on 2023-12-05, 21:13:

The parts might be repainted, but you got what the markings state.

Why do you think that? What would be the purpose in remarking the chips with the same part number and speed?

Why would 50 ns NPN chips work at 80 MHz MCLK, but the 45 ns V53C chips would not?

CoffeeOne, were your original V53C chips able to work at 80 MHz?

Hi, I already mentioned it. YES.
There are also V53C104FK60 and V53C104FK70.
The HP are the "high performance" series.
So I bet, the chips you got were V53C104FK70.

CoffeeOne wrote on 2023-12-05, 21:46:

There are also V53C104FK60 and V53C104FK70.
The HP are the "high performance" series.
So I bet, the chips you got were V53C104FK70.

I was thinking something similar. I suppose I should try another set of 8 from my shipment to ensure that I didn't get just one slow chip in the mix.

I have since independently tested 3 of the fake V53C104HP45 chips, in the configuration shown below:

From the 3 separate tests, booting into Windows 95c at 1280x1024x256c-60Hz, and 80 MHz MCLK, they all looked as follows:

Next, I grabbed a 60 ns module, HYB514256B-60, and I do not see any artefacts, as shown:

Next, I tried a 70 ns chip, V53C104BP70, as shown:

The artefacts look like this:

Next, I tried an 80 ns chip, V53C104AP80, as shown:

The artefacts get worse, and look like this:

The artefacting of the fake V53C104HP45 looks similar to that of the original V53C104BP70. Without measuring the actual response times, it looks like the fake V53C104HP45 may be 70 ns remarked chips. What do you guys think?

Another discovery was that 60 ns is the approximate minimum DRAM speed for running MCLK at 80 MHz.

This is actually common with vlb cards.
You are not a victim of some fraud. Well, maybe you are, i cannot say, but had plenty of exactly this kind of problems.

In the other hand these are almost nonexistent with pci cards.

Can remember few cases with isa cards, but not common for sure.

Does anybody know something about the ALG1201 RAMDAC? It is used on advanced logic graphics card only apparently, so I guess it is not ATT20C490 or ATT20C491 compatible. Or Ark1491 compatible.
I cannot find a specification for the ALG1201.

Possibly compatible are Sierra Semiconductor SC15025CV and SC15026CV, for which 125MHz variants exist. But the chips are hard to find anyway, so maybe not worth the effort to search for datasheets.